Automation index

How much of a ML engineer’s job can AI already do?

Training a baseline and a notebook is fast. Shipping it, watching drift, and owning the miss is not.

This percentage does not predict how many people will lose their jobs.

Task comparison

IS WORK OVER?

29%

AI can already do 29% of this job.

#64 / 623

ML engineer

Task weights are our estimates based on O*NET frequency data, not time shares published by O*NET.

The time saving for each task is estimated separately.

Task	Share	AI	Points
Train models, features and notebooks. Estimate No public source provides a precise figure. This is our estimate.	22%	48%	+10.6
Evaluate, document and compare runs. Estimate No public source provides a precise figure. This is our estimate.	18%	42%	+7.6
Deploy, monitor and retrain. Estimate No public source provides a precise figure. This is our estimate.	16%	22%	+3.5
Talk to product about what a good error is. Estimate No public source provides a precise figure. This is our estimate.	16%	15%	+2.4
Dirty data, bias and model limits. Estimate No public source provides a precise figure. This is our estimate.	14%	18%	+2.5
On-call, incidents and rollback. Estimate No public source provides a precise figure. This is our estimate.	14%	16%	+2.2
Total	100%		29%

Modelled workweek: 100% · Public evidence: 0% · Estimate: 100% · Confidence: Low

IS WORK OVER?

28%

#92 / 623

Mobile developer

Task weights are our estimates based on O*NET frequency data, not time shares published by O*NET.

The time saving for each task is estimated separately.

Task	Share	AI	Points
Screens, API and offline state. Estimate No public source provides a precise figure. This is our estimate.	18%	52%	+9.4
Store review, crashlytics and ANRs. Estimate No public source provides a precise figure. This is our estimate.	16%	40%	+6.4
Real device, the notch and fragmented Android. Estimate No public source provides a precise figure. This is our estimate.	16%	22%	+3.5
CI, tests and the week's release. Estimate No public source provides a precise figure. This is our estimate.	16%	28%	+4.5
Product, the redesign and not breaking what people use. Estimate No public source provides a precise figure. This is our estimate.	18%	14%	+2.5
That it does not crash, and not ignoring the one-star. Estimate No public source provides a precise figure. This is our estimate.	16%	12%	+1.9
Total	100%		28%

Modelled workweek: 100% · Public evidence: 0% · Estimate: 100% · Confidence: Low